A Novel Experimental Technique for Recovering Camera Response Function for High Dynamic Range Imaging
Salvador Garcia Bernal, Yuanhao Gong, Bozhi Liu, Guoping Qiu · 2018
Recovering the camera's response function has wide applica- tion in image processing and computer vision. Traditional computational approach using multiple differently exposed photographs is based on solving a constrained linear least square problem and can only recover the function up to a scaling factor. In this paper, we present a novel experiment based technique to recover the true response curves of spe- cific sensors. Our method first measures the average pixel intensities of a neutrally-grey cardboard for the full range of the shutter speed and gain control values of the sensor. Af- ter a simple data smoothing procedure, the camera's response function is recovered through simple table look-up. We have implemented our technique on different kinds of sensors and show that it works well. The new technique recover the actual range of the sensor and is more accurate than the computa- tional technique. The procedure is easy to set up and simple to implement which will be useful for developing applications such as High Dynamic Ranging (HDR) imaging.